Local Linearity and Double Descent in Catastrophic Overfitting

• URL: http://arxiv.org/abs/2111.10754v1
• Date: Sun, 21 Nov 2021 07:11:21 GMT
• Title: Local Linearity and Double Descent in Catastrophic Overfitting
• Abstract summary: Catastrophic overfitting is a phenomenon observed during Adversarial Training (AT) with the Fast Gradient Sign Method (FGSM) We show that maintaining high local linearity might be $textitsufficient$ to prevent catastrophic overfitting but is not $textitnecessary.
• Score: 0.0
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: Catastrophic overfitting is a phenomenon observed during Adversarial Training (AT) with the Fast Gradient Sign Method (FGSM) where the test robustness steeply declines over just one epoch in the training stage. Prior work has attributed this loss in robustness to a sharp decrease in $\textit{local linearity}$ of the neural network with respect to the input space, and has demonstrated that introducing a local linearity measure as a regularization term prevents catastrophic overfitting. Using a simple neural network architecture, we experimentally demonstrate that maintaining high local linearity might be $\textit{sufficient}$ to prevent catastrophic overfitting but is not $\textit{necessary.}$ Further, inspired by Parseval networks, we introduce a regularization term to AT with FGSM to make the weight matrices of the network orthogonal and study the connection between orthogonality of the network weights and local linearity. Lastly, we identify the $\textit{double descent}$ phenomenon during the adversarial training process.

Related papers

• Find A Winning Sign: Sign Is All We Need to Win the Lottery [52.63674911541416]
  We show that a sparse network trained by an existing IP method can retain its basin of attraction if its parameter signs and normalization layer parameters are preserved. To take a step closer to finding a winning ticket, we alleviate the reliance on normalization layer parameters by preventing high error barriers along the linear path between the sparse network trained by our method and its counterpart with normalization layer parameters.
  arXiv  Detail & Related papers  (2025-04-07T09:30:38Z)
• Efficient local linearity regularization to overcome catastrophic overfitting [59.463867084204566]
  Catastrophic overfitting (CO) in single-step adversarial training results in abrupt drops in the adversarial test accuracy (even down to 0%) We introduce a regularization term, called ELLE, to mitigate CO effectively and efficiently in classical AT evaluations.
  arXiv  Detail & Related papers  (2024-01-21T22:55:26Z)
• Grokking as the Transition from Lazy to Rich Training Dynamics [35.186196991224286]
  grokking occurs when the train loss of a neural network decreases much earlier than its test loss. Key determinants of grokking are the rate of feature learning and the alignment of the initial features with the target function.
  arXiv  Detail & Related papers  (2023-10-09T19:33:21Z)
• Adversarial Examples Exist in Two-Layer ReLU Networks for Low Dimensional Linear Subspaces [24.43191276129614]
  We show that standard methods lead to non-robust neural networks. We show that decreasing the scale of the training algorithm, or adding $L$ regularization, can make the trained network more robust to adversarial perturbations.
  arXiv  Detail & Related papers  (2023-03-01T19:10:05Z)
• Implicit Bias in Leaky ReLU Networks Trained on High-Dimensional Data [63.34506218832164]
  In this work, we investigate the implicit bias of gradient flow and gradient descent in two-layer fully-connected neural networks with ReLU activations. For gradient flow, we leverage recent work on the implicit bias for homogeneous neural networks to show that leakyally, gradient flow produces a neural network with rank at most two. For gradient descent, provided the random variance is small enough, we show that a single step of gradient descent suffices to drastically reduce the rank of the network, and that the rank remains small throughout training.
  arXiv  Detail & Related papers  (2022-10-13T15:09:54Z)
• On the Effective Number of Linear Regions in Shallow Univariate ReLU Networks: Convergence Guarantees and Implicit Bias [50.84569563188485]
  We show that gradient flow converges in direction when labels are determined by the sign of a target network with $r$ neurons. Our result may already hold for mild over- parameterization, where the width is $tildemathcalO(r)$ and independent of the sample size.
  arXiv  Detail & Related papers  (2022-05-18T16:57:10Z)
• Benign Overfitting in Two-layer Convolutional Neural Networks [90.75603889605043]
  We study the benign overfitting phenomenon in training a two-layer convolutional neural network (CNN) We show that when the signal-to-noise ratio satisfies a certain condition, a two-layer CNN trained by gradient descent can achieve arbitrarily small training and test loss. On the other hand, when this condition does not hold, overfitting becomes harmful and the obtained CNN can only achieve constant level test loss.
  arXiv  Detail & Related papers  (2022-02-14T07:45:51Z)
• Training invariances and the low-rank phenomenon: beyond linear networks [44.02161831977037]
  We show that when one trains a deep linear network with logistic or exponential loss on linearly separable data, the weights converge to rank-$1$ matrices. This is the first time a low-rank phenomenon is proven rigorously for nonlinear ReLU-activated feedforward networks. Our proof relies on a specific decomposition of the network into a multilinear function and another ReLU network whose weights are constant under a certain parameter directional convergence.
  arXiv  Detail & Related papers  (2022-01-28T07:31:19Z)
• Robustness Certificates for Implicit Neural Networks: A Mixed Monotone Contractive Approach [60.67748036747221]
  Implicit neural networks offer competitive performance and reduced memory consumption. They can remain brittle with respect to input adversarial perturbations. This paper proposes a theoretical and computational framework for robustness verification of implicit neural networks.
  arXiv  Detail & Related papers  (2021-12-10T03:08:55Z)
• DL-Reg: A Deep Learning Regularization Technique using Linear Regression [4.1359299555083595]
  This paper proposes a novel deep learning regularization method named as DL-Reg. It carefully reduces the nonlinearity of deep networks to a certain extent by explicitly enforcing the network to behave as much linear as possible. The performance of DL-Reg is evaluated by training state-of-the-art deep network models on several benchmark datasets.
  arXiv  Detail & Related papers  (2020-10-31T21:53:24Z)
• A Unifying View on Implicit Bias in Training Linear Neural Networks [31.65006970108761]
  We study the implicit bias of gradient flow (i.e., gradient descent with infinitesimal step size) on linear neural network training. We propose a tensor formulation of neural networks that includes fully-connected, diagonal, and convolutional networks as special cases.
  arXiv  Detail & Related papers  (2020-10-06T06:08:35Z)
• Structure Learning in Inverse Ising Problems Using $\ell_2$-Regularized Linear Estimator [8.89493507314525]
  We show that despite the model mismatch, one can perfectly identify the network structure using naive linear regression without regularization. We propose a two-stage estimator: In the first stage, the ridge regression is used and the estimates are pruned by a relatively small threshold. This estimator with the appropriate regularization coefficient and thresholds is shown to achieve the perfect identification of the network structure even in $0M/N1$.
  arXiv  Detail & Related papers  (2020-08-19T09:11:33Z)
• Feature Purification: How Adversarial Training Performs Robust Deep Learning [66.05472746340142]
  We show a principle that we call Feature Purification, where we show one of the causes of the existence of adversarial examples is the accumulation of certain small dense mixtures in the hidden weights during the training process of a neural network. We present both experiments on the CIFAR-10 dataset to illustrate this principle, and a theoretical result proving that for certain natural classification tasks, training a two-layer neural network with ReLU activation using randomly gradient descent indeed this principle.
  arXiv  Detail & Related papers  (2020-05-20T16:56:08Z)
• Revisiting Initialization of Neural Networks [72.24615341588846]
  We propose a rigorous estimation of the global curvature of weights across layers by approximating and controlling the norm of their Hessian matrix. Our experiments on Word2Vec and the MNIST/CIFAR image classification tasks confirm that tracking the Hessian norm is a useful diagnostic tool.
  arXiv  Detail & Related papers  (2020-04-20T18:12:56Z)
• A Generalized Neural Tangent Kernel Analysis for Two-layer Neural Networks [87.23360438947114]
  We show that noisy gradient descent with weight decay can still exhibit a " Kernel-like" behavior. This implies that the training loss converges linearly up to a certain accuracy. We also establish a novel generalization error bound for two-layer neural networks trained by noisy gradient descent with weight decay.
  arXiv  Detail & Related papers  (2020-02-10T18:56:15Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.